Downloads · 30 days
18
4% of all-time downloads
anuragshas/wav2vec2-large-xls-r-300m-mr
wav2vec2-large-xls-r-300m-mr is a automatic speech recognition model from anuragshas. Use it when you need speech turned into text. It is set up for transformers. The card lists the license as apache-2.0.
should probably proofread and complete it, then remove this comment. --
Downloads · 30 days
18
4% of all-time downloads
All-time downloads
443
Public
Repo size
5.6 GB
Likes
0
Public
Click a slice to open those files.
.bin3.1 GB · 99%
From the Hugging Face model README
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the common_voice dataset. It achieves the following results on the evaluation set:
More information needed
More information needed
More information needed
The following hyperparameters were used during training:
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 3.7378 | 18.18 | 400 | 3.5047 | 1.0 |
| 3.1707 | 36.36 | 800 | 2.6166 | 0.9912 |
| 1.4942 | 54.55 | 1200 | 0.5778 | 0.6927 |
| 1.2058 | 72.73 | 1600 | 0.5168 | 0.6362 |
| 1.0558 | 90.91 | 2000 | 0.5105 | 0.6069 |
| 0.9488 | 109.09 | 2400 | 0.5151 | 0.6089 |
| 0.8588 | 127.27 | 2800 | 0.5157 | 0.5989 |
| 0.7991 | 145.45 | 3200 | 0.5179 | 0.5740 |
| 0.7545 | 163.64 | 3600 | 0.5348 | 0.5740 |
| 0.7144 | 181.82 | 4000 | 0.5518 | 0.5724 |
| 0.7041 | 200.0 | 4400 | 0.5479 | 0.5740 |
mozilla-foundation/common_voice_8_0 with split testpython eval.py --model_id anuragshas/wav2vec2-large-xls-r-300m-mr --dataset mozilla-foundation/common_voice_8_0 --config mr --split test
import torch
from datasets import load_dataset
from transformers import AutoModelForCTC, AutoProcessor
import torchaudio.functional as F
model_id = "anuragshas/wav2vec2-large-xls-r-300m-mr"
sample_iter = iter(load_dataset("mozilla-foundation/common_voice_8_0", "mr", split="test", streaming=True, use_auth_token=True))
sample = next(sample_iter)
resampled_audio = F.resample(torch.tensor(sample["audio"]["array"]), 48_000, 16_000).numpy()
model = AutoModelForCTC.from_pretrained(model_id)
processor = AutoProcessor.from_pretrained(model_id)
input_values = processor(resampled_audio, return_tensors="pt").input_values
with torch.no_grad():
logits = model(input_values).logits
transcription = processor.batch_decode(logits.numpy()).text
# => "या पानास लेखाचे स्वरूप यायला हावे"
| Without LM | With LM (run ./eval.py) |
|---|---|
| 49.177 | 32.811 |